Handle quoted fields with embedded commas correctly
CSV fields that contain the delimiter character must be wrapped in double quotes: "Smith, John" is a single field containing a comma. The converter handles standard RFC 4180 CSV quoting automatically — quoted fields with embedded commas, quotes escaped as double-quotes within quoted fields, and multi-line fields wrapped in quotes. If your data has unusual quoting, switch to a different delimiter (semicolon or tab) that does not appear in your data values.
Check encoding before converting
CSV files from Excel often use Windows-1252 or Latin-1 encoding rather than UTF-8. Pasting such content directly may produce garbled characters for accented letters, currency symbols, and non-Latin scripts. If you see characters like é instead of é, re-open the CSV in a text editor, save as UTF-8, then paste. Most modern data tools export UTF-8 by default.
For large files, use the tab delimiter export from Excel
Excel's CSV export sometimes mishandles fields with embedded commas, newlines, or special characters. Export as "Tab Delimited Text" instead, then switch the converter's delimiter to Tab. Tab characters almost never appear in data values, making tab-delimited exports more reliable than comma-delimited for complex data.
Flatten nested JSON before converting to CSV
The JSON-to-CSV converter handles flat arrays of objects — where each value is a primitive (string, number, boolean). Nested objects ({"address": {"city": "London", "country": "UK"}}) cannot be directly represented as CSV columns. Flatten nested structures first: manually or programmatically transform {"address.city": "London", "address.country": "UK"} before converting. The preview table will show any nested objects as [object Object] which is a signal that flattening is needed.
Use type inference for API consumption
When converting CSV data that will be consumed by an API or JavaScript application, always enable type inference. Numeric IDs, ages, prices, and counts should be numbers in JSON, not strings — numeric comparisons, sorting, and arithmetic will all work incorrectly on string "42" vs numeric 42. The type inference converts unambiguous numeric and boolean values automatically.